Abstract
Facial expression recognition was investigated in 20 males with high functioning autism (HFA) or Asperger syndrome (AS), compared to typically developing individuals matched for chronological age (TD CA group) and verbal and non-verbal ability (TD V/NV group). This was the first study to employ a visual search, “face in the crowd” paradigm with a HFA/AS group, which explored responses to numerous facial expressions using real-face stimuli. Results showed slower response times for processing fear, anger and sad expressions in the HFA/AS group, relative to the TD CA group, but not the TD V/NV group. Reponses to happy, disgust and surprise expressions showed no group differences. Results are discussed with reference to the amygdala theory of autism.
. Results
2.1. Analysis of response times (RT)
RTs of correct response only were analysed. One participant from the ASD group responded incorrectly to every trial with fear as a target. This empty cell was replaced with the mean RT for the remaining 19 HFA/AS participants for this expression. Results are shown in Fig. 4.
Response times to presentations of each target emotion: mean (s.e.).
Fig. 4.
Response times to presentations of each target emotion: mean (s.e.).
Figure options
ANOVA was carried out with target emotion (6 levels; anger, disgust, fear, happiness, sadness, surprise) as the within-subjects factor, and group (3 levels; HFA/AS, TD CA and TD V/NV) as the between-subjects factor. This showed a significant main effect of target emotion (F(3.64, 207.53) = 55.91, p < 0.001, partialη2 = 0.50). Pair-wise comparison (Bonferroni corrected) showed that this was due to significantly faster reaction times for happy faces than for all other emotional expressions (p < 0.05 for all), and significantly slower recognition of sad, fearful and angry faces than surprise, disgust and happy emotional expressions (p < 0.05 for all). The main effect of group was also significant (F(2, 57) = 4.52, p = 0.02, partialη2 = 0.14). Post hoc Tukey analysis indicated that the HFA/AS participants were significantly slower than TD CA participants (p < 0.05), but not the TD V/NV control group (p > 0.05) overall. The responses of the TD CA and TD V/NV groups did not differ from one another (p > 0.05).
The interaction between group and target emotion was also significant (F(7.28, 207.53) = 2.45, p = 0.01, partialη2 = 0.08). Further exploration revealed that the main effect of group above was driven by a subset of emotional expressions; there was no significant difference between groups in the perception of disgusted faces (F(2, 57) = 2.30, p > 0.05), happy faces (F > 1) or surprised faces (F < 1). Significant differences between groups were found for angry faces (F(2, 57) = 3.79, p < 0.05), fearful faces (F(2, 57) = 6.84, p < 0.01) and sad faces (F(2, 57) = 4.60, p < 0.05). Further Tukey post hoc analysis for angry, fearful and sad faces showed similar patterns in that the HFA/AS showed significantly longer RTs than the TD CA group (p < 0.05), but not the TD V/NV group (p > 0.05), and that the RTs of the TD CA and TD V/NV groups did not differ (p > 0.05).
2.2. Analysis of errors
As visual search tasks are designed to produce no or very low numbers of errors, RT analyses are typically most informative. However, as errors were made, performance was compared to ceiling to determine whether analysis of error rates was justified. In the ASD group errors were all significantly different from zero (p < 0.05) with the exception of disgust (p = 0.10) and happy (p = 0.08). In the TD CA group performance was significantly different from zero in each emotion (p < 0.05 for all). In the TD V/NV group, performance for all emotions was significantly different from zero (p < 0.05 for all) except for happy (p = 0.08). Results are shown in Fig. 5.
Number of errors to presentations of each target emotion: mean (s.e.).
Fig. 5.
Number of errors to presentations of each target emotion: mean (s.e.).
Figure options
As ceiling effects were infrequent, ANOVA was carried out on error rates with target emotion (6 levels; anger, disgust, fear, happiness, sadness and surprise) as the within-subjects factor, and group (3 levels; HFA/AS, TD CA and TD V/NV) as the between-subjects factor. Results revealed no main effect of group (F < 1). There was, however, a main effect of emotion (F (3.22, 183.37) = 11.81, p < 0.001, partialη2 = 0.17). Post hoc analysis (Bonferroni corrected) revealed the least errors for responses to happy (significantly lower than anger, fear, sad and surprise, p < 0.05 for all) and disgusted faces (significantly lower than anger, fear and surprise, p < 0.05 for all) and the most errors for anger (significantly higher than happy and disgust, p < 0.05 for both) and fear (significantly higher than happy, disgust and sad, p < 0.05 for all), with intermediate and similar (p > 0.05) error rates for sad and surprise. No significant emotion × group interaction was found (F (6.43, 183.37) = 1.06, p = 0.39, partialη2 = 0.04).
2.3. Correlational analysis
Mean RT and mean error variables were calculated for each participant across the six emotional expression variables, and their relationship to chronological age (CA), level of verbal ability (measured by the BPVS) and level of non-verbal ability (measured by the RCPM) determined through correlational analysis. In order to take a conservative approach, two-tailed significance values are reported. Mean RT scores were negatively associated with CA only for the ASD and TD V/NV groups (ASD: R = −0.52, p = 0.02; TD V/NV: R = −0.58, p = 0.01), with no significant associations for the TD CA group (R = −0.36, p = 0.12). Mean error scores were negatively associated with RCPM for all three groups, although only marginally so for the ASD group (TD CA: R = −0.52, p = 0.02; TD V/NV: R = −0.51, p = 0.02 ASD: R = −0.39, p = 0.09), and also with BPVS for the TD V/NV group (R = −0.68, p = 0.001). There were no other significant associations
